This application is related to subject matter disclosed in:
A1) U.S. provisional Application No. 60/257,218 entitled “Waveguides and resonators for integrated optical devices and methods of fabrication and use thereof” filed Dec. 21, 2000 in the name of Oskar J. Painter, said provisional application being hereby incorporated by reference in its entirety as if fully set forth herein;
A2) U.S. provisional Application No. 60/301,519 entitled “Waveguide-fiber Mach-Zender interferometer and methods of fabrication and use thereof” filed Jun. 27, 2001 in the names of Oskar J. Painter, David W. Vernooy, and Kerry J. Vahala, said provisional application being hereby incorporated by reference in its entirety as if fully set forth herein;
A3) U.S. provisional Application No. 60/322,272 entitled “Fiber-optic-taper probe for characterizing transversely-optically-coupled waveguides and resonators” filed Sep. 13, 2001 in the name of David W. Vernooy, said provisional application being hereby incorporated by reference in its entirety as if fully set forth herein;
A4) U.S. Pat. No. 5,032,219 entitled “Method for improving the planarity of etched mirror facets” issued Jul. 16, 1991 in the names of Peter. L. Buchman, Peter Vettiger, Otto Voegeli, and David J. Webb, said patent being hereby incorporated by reference in its entirety as if fully set forth herein;
A5) U.S. Pat. No. 5,103,493 entitled “Improved planar etched mirror facets” issued Apr. 7, 1992 in the names of Peter. L. Buchman, Peter Vettiger, Otto Voegeli, and David J. Webb, said patent being hereby incorporated by reference in its entirety as if fully set forth herein;
A6) U.S. Pat. No. 5,177,031 entitled “Method of passivating etched mirror facets of semiconductor laser diodes” issued Jan. 5, 1993 in the names of Peter. L. Buchman, David J. Webb, and Peter Vettiger, said patent being hereby incorporated by reference in its entirety as if fully set forth herein;
A7) U.S. Pat. No. 5,259,049 entitled “Self-aligned optical waveguide to laser structure and method of making the same” issued Nov. 2, 1993 in the names of Gian-Luca Bona, Fritz Gfeller, Heinz Jaeckel, and David J. Webb, said patent being hereby incorporated by reference in its entirety as if fully set forth herein;
A8) U.S. provisional App. No. 60/334,705 entitled “Integrated end-coupled transverse-optical-coupling apparatus and methods” filed Oct. 30, 2001 in the names of Henry A. Blauvelt, Kerry J. Vahala, Peter C. Sercel, Oskar J. Painter, and Guido Hunziker, said application being hereby incorporated by reference in its entirety as if fully set forth herein;
A9) U.S. provisional App. No. 60/333,236 entitled “Alignment apparatus and methods for transverse optical coupling” filed Nov. 23, 2001 in the names of Charles I. Grosjean, Guido Hunziker, Paul M. Bridger, and Oskar J. Painter, said application being hereby incorporated by reference in its entirety as if fully set forth herein;
A10) U.S. non-provisional application Ser. No. 10/037,966 entitled “Multi-layer dispersion-engineered waveguides and resonators” filed Dec. 21, 2001 in the names of Oskar J. Painter, David W. Vernooy, and Kerry J. Vahala, said application being hereby incorporated by reference in its entirety as if fully set forth herein; and
A11) U.S. provisional App. No. 60/360,261 entitled “Alignment-insensitive optical junction apparatus and methods employing adiabatic optical power transfer” filed Feb. 27, 2002 in the names of Henry A. Blauvelt, Kerry J. Vahala, David W. Vernooy, and Joel S. Paslaski, said provisional application being hereby incorporated by reference as if fully set forth herein.
This application is also related to subject matter disclosed in the following publications, each of said publications being hereby incorporated by reference in its entirety as if fully set forth herein:
P1) Y. P. Li and C. H. Henry, Silicon Optical Bench Waveguide Technology, in Optical Fiber Telecommunications, IIIb, I. P. Kaminow and T. L. Koch eds., Academic Press, 1997;
P2) T. Ramadan, R. Scarmozzino, and R Osgood “Adiabatic Couplers: Design Rules and Optimization” IEEE J. Lightwave Tech., v16, No. 2, pp 277-283, (1998);
P3) D. G. Dalgoutte, R. B. Smith, G. Achutaramayya, and J. H. Harris, “Externally mounted fibers for integrated optics interconnections”, Appl. Optics Vol. 14, No. 8, pp 1860-1865 (1975); and
P4) Y. Shani, C. H. Henry, R. C. Kistler, R. F. Kazarinov, and K. J. Orlowsky, “Integrated optic adiabatic devices on silicon”, IEEE J. Quant. Elec., Vol. 27, No. 3, pp556-566 (1991).
A fundamental problem in the field of optical telecommunications is attaining efficient and cost-effective transfer of optical signal power between assembled optical components. One particularly significant example is achieving optical signal power transfer between an active or passive optical device and a low-loss transmission optical waveguide, including optical fibers and/or planar waveguide circuits. Examples of active optical devices may include but are not limited to semiconductor lasers, electro-absorption modulators, electro-absorption modulated lasers, electro-optic modulators, semiconductor optical amplifiers, photodiodes or other photodetectors, N×N optical switches, and so forth. Examples of passive devices may include but are not limited to wavelength division multiplexers/de-multiplexers, wavelength division slicers/interleavers, wavelength division add/drop filters, other optical filters, splitters/combiners, interferometers, phase shifters, dispersion compensators, fixed or variable optical attenuators, and so forth. Such optical devices often involve generation of, interaction with, and/or manipulation of optical modes that are typically small (particularly in semiconductor-based devices), often on the order of just a few microns across and sometimes less than 1 micron high. This interacting mode size is typically much smaller than an optical mode size supported by a single-mode optical fiber or a planar lightwave circuit (generally about ten microns across). End-coupling of an optical fiber or planar waveguide circuit to an optical device is therefore often inefficient (around 5-15%) due to spatial mode mismatch, yielding devices having undesirably large insertion losses. Prior art methodologies exist for achieving higher end-coupling efficiencies, but these require expensive components for achieving better mode-matching (aspheric lenses and the like), and also require high-precision active alignment of the optical components and the optical device (required tolerances may be as small as 0.1 μm, and must typically be achieved on an individual device basis).
Prior art methodologies exist for low-cost end-coupled optical assembly (such as methodologies based on silicon optical bench technologies, for example). However, these low-cost solutions generally suffer from low optical power transfer efficiency between an optical device and an optical fiber or other waveguide, for the reasons set forth hereinabove.
Optical power transfer by end-coupling (equivalently, end-fire coupling or end-transfer) is characterized by positioning of the optical components in an end-to-end geometry substantially along the direction of propagation of the optical signal power to be transferred. At the optical junction thus formed, optical power propagates out through an end-face of one optical component and in through an end-face of another optical component. Alternatively, optical power transfer may be achieved by so-called transverse-coupling (equivalently, transverse-transfer), in which the optical components are positioned in a side-by-side geometry relative to the direction of propagation of the optical signal power. At the optical junction formed by transverse-coupling, there is typically at least one segment of the junction with optical power propagating along both components simultaneously.
Efficient end-transfer between optical components requires that optical modes in the respective components be substantially spatial-mode matched. Transverse-transfer of optical power between an optical device and a transmission optical waveguide provides an alternative to end-transfer for transferring optical signal power between an optical device and a transmission waveguide (through a taper segment of an optical fiber or through a suitably adapted portion of a planar waveguide, for example). In particular, the requirement for spatial-mode matching is eliminated; transverse-transfer of optical power may be achieved between optical modes of differing spatial-mode size and/or shape.
Transverse-transfer (also referred to as transverse coupling, transverse optical coupling, evanescent optical coupling, evanescent coupling, directional optical coupling, directional coupling) is discussed at length in several of the prior patent applications cited hereinabove, and the entire discussion need not be repeated herein. Transverse-transfer may be readily described in terms of optical modes characteristic of the separate optical waveguides (or other optical components) transitioning to the optical modes characteristic of a coupled-waveguide optical system. These latter modes are referred to herein as the “system modes” or “coupled-system modes”, while the former modes are referred to herein as the “isolated modes” or “isolated-waveguide modes”. Efficient transfer of optical signal power between optical waveguides by transverse-coupling may be achieved in one of several operating regimes. Two such regimes discussed herein are so-called mode-interference coupling and so-called adiabatic optical power transfer.
In so-called mode-interference coupling (described in several of the above-cited references, particularly A8 and A10, and referred to therein simply as transverse optical coupling), optical signal power entering a junction region from one waveguide is divided between two guided system modes. Ideally, this transition into the junction region is configured so that the isolated mode is very nearly a linear superposition of the two lowest order system modes. This condition results in minimal power loss to higher order system modes (and/or radiation modes) as optical signal power enters the junction region. The two system modes propagate through the junction region along the waveguides with differing propagation constants (designated as β+ and β− for the two lowest-order system modes). Upon reaching the end of the junction region, optical signal power is divided into the two waveguides according to the relative phase of the two system modes. Once again, to minimize loss to higher-order and/or radiative modes, the isolated modes should substantially resemble linear superpositions of the two system modes. Since this is typically the case in practical devices, and presents a reasonable approximation even when it is not the case, it is usually possible to describe the characteristics of the junction region in terms of properties of the isolated modes, and such a description shall be used hereinafter. In particular, the degree of optical signal power transfer via mode-interference coupling is determined by the degree of transverse overlap between the isolated-waveguide modes, (characterized by a coupling coefficient κ), by the propagation distance over which the modes overlap (i.e., junction region length or interaction length L), and by the degree of modal index mismatch (characterized by Δβ=β1−β2, the β's being the propagation constants for the respective isolated-waveguide modes). In mode-interference coupling, κ, β1, β2 are typically assumed to remain substantially constant over the length L of the junction region. Transfer of optical power between the mode-interference-coupled waveguides is given by (neglecting the effects of optical losses):                     |                              E            2                    ⁢                      (            L            )                          ⁢                  |          2                            |                              E            1                    ⁢                      (            0            )                          ⁢                  |          2                      =                            |          κ          ⁢                      |            2                                    q          2                    ⁢                        sin          2                ⁡                  (                                                    q                                            L                                              )                                q      2        =          |      κ      ⁢              |        2            ⁢                        +                      1            4                          ⁢                              Δβ            2                    .                    where the following definitions apply:                E1,2(z) amplitudes of the coupled fields;        β1,2 propagation constants of the coupled fields;        κ coupling amplitude resulting from spatial overlap of the fields;        z longitudinal propagation distance coordinateAn incident field of amplitude E1 that is spatially confined to a first optical waveguide before the junction region will transfer to the other optical element with a resultant field amplitude E2(L) at z=L (where we define z=0 as the start of the junction region and z=L as the end of the junction region). Optical power transfer as a function of the junction region length L is therefore oscillatory with a characteristic period or “beat length” that depends on κ and Δβ. This may be thought of as a manifestation of the interference between the system modes excited within the junction region, both of which carry optical signal power. Greater coupling amplitude κ and/or greater modal-index mismatch Δβ will reduce the beat length. The absolute magnitude of the oscillatory power transfer decreases with increasing modal-index mismatch, with substantially complete transfer of optical power back and forth between the optical elements when Δβ is near zero. A particular degree of optical power transfer from one waveguide to the other may be achieved by configuring the junction region with the length L to achieve the desired transfer fraction for a given Δβ and κ.        
To understand the distinction between mode interference coupling and adiabatic power transfer, it is first necessary to understand the meaning of the adiabatic condition within the general context of an optical waveguide. Two examples are presented for illustration. Consider first a single mode waveguide that is tapered over some segment of its length so as to modify both the transverse extent and the propagation constant of the guided mode. Tapering of a waveguide supporting even a single mode induces coupling to radiation modes. However, provided that the tapering is sufficiently gradual so that this radiative loss is weak (i.e., adiabatic tapering), it still makes sense to consider the optical power traversing the tapered waveguide as representing a single mode, albeit one whose properties have a longitudinal position dependence (i.e. z-dependence) as it traverses the tapered waveguide segment. Provided the adiabatic condition is satisfied (i.e., tapering is slow enough to render coupling to other modes minimal or below an operationally acceptable level), it is possible to describe the mode using longitudinally varying quantities such as a z-dependent propagation “constant” β(z).
As a second example, the properties of a waveguide could be varied along the longitudinal propagation direction so that the waveguide at one position supports a single transverse mode while at another position supports two or more transverse modes. In this example, adiabatic variation of the waveguide properties would result in negligible (or operationally acceptable) coupling to these other modes so that once again it is possible to think of the single “mode” as being preserved as it propagates along the waveguide, albeit as a mode whose properties such as its propagation “constant” β and/or its transverse spatial profile acquire a dependence on longitudinal position z along the waveguide.
This approximate way of considering optical modes subject to an adiabatic variation along the longitudinal or propagation direction is an important concept for understanding the operation of adiabatic power transfer devices. It is important to note that the term “mode” acquires a slightly more general meaning in the context of waveguides and junctions that satisfy an adiabatic condition. In particular, to the extent that coupling to other modes is minimal or remains at or below some operationally acceptable level, the terms “mode” and/or “optical mode” shall be used herein even if spatial, temporal, polarization, and/or other properties might evolve as the mode propagates along a waveguide whose properties vary longitudinally in an adiabatic fashion. This more general interpretation of modes is distinct from the more conventional use of the term “mode” which may typically imply preservation of certain modal properties, such as propagation constant β, transverse spatial profile, polarization state, and so on, as the mode propagates along a substantially longitudinally invariant waveguide.
For adiabatic optical power transfer, two isolated modes a1(z) and a2(z) characteristic of the isolated waveguides begin to experience weak coupling as they enter the junction region. Under the adiabatic condition this weak coupling may be characterized by a coupling coefficient κ(Z) and modal-index mismatch Δβ(z)=β1(Z)−β2(Z). The resulting system modes will substantially resemble the superposition modes a+(z) and a−(z) of the coupled-waveguide system given by       a    ±    =                                          λ            ±                                                              κ                2                            +                              λ                ±                2                                                    ⁢                  a          1                    +                        κ                                                    κ                2                            +                              λ                ±                2                                                    ⁢                  a          2                ⁢                                   ⁢        w        ⁢                                   ⁢        h        ⁢                                   ⁢        e        ⁢                                   ⁢        r        ⁢                                   ⁢        e        ⁢                                   ⁢                  λ          ±                      =                  (                  Δβ          2                )            ±                                    κ            2                    +                                    (                              Δβ                2                            )                        2                              where all quantities are z-dependent. For purposes of the present discussion, the terms “superposition modes” and “system modes” may be used interchangeably, even though the system modes may not resemble the superposition modes throughout the junction region. At the beginning of the junction region (i.e., z=0), superposition mode a+ preferably closely resembles only one of the isolated-waveguide modes a1 or a2, while mode a− resembles the other. For example, in the limit of |Δβ|>>|κ| (i.e., strongly modal-index mis-matched),             a      +        ≈                  a        1            +                        κ          Δβ                ⁢                  a          2                      ≈                  a        1            ⁢                           ⁢      a      ⁢                           ⁢      n      ⁢                           ⁢      d      ⁢                           ⁢              a        -              ≈                  a        2            -                        κ          Δβ                ⁢                  a          1                      ≈          a      2        ,meaning each superposition mode is predominantly associated with a single isolated-waveguide mode in this limit (i.e., a+a1 and a−a2). For adiabatic optical coupling, preferably |Δβ|>>|κ| for the isolated-waveguide modes at z=0. Under this input termination condition, the superposition modes (and hence also the system modes) substantially resemble the isolated-waveguide modes, and optical signal power entering the junction region in a first waveguide is transferred predominantly (even exclusively) into the corresponding system mode. The junction region is configured so that |Δβ| (for the isolated-waveguide modes) initially decreases along the junction region. The coefficient κ may also vary along the junction region, preferably reaching a maximum absolute value within the junction region. As evident from the equations defining the superposition modes given above, the variation of Δβ and/or κ results in evolution of the superposition modes (more precisely, the system modes) along the length of the junction region. As described above, the adiabatic condition requires that the variation of Δβ and/or κ must be sufficiently gradual so that transfer of optical power between system modes and/or between a system mode and other optical modes (guided or otherwise) remains at or below some operationally acceptable level. This criterion is equivalent to the adiabatic condition described in reference P2. In particular, any change in waveguide spacing, transverse dimensions, modal and/or material index, or other properties (before, within, and/or after the junction region) must be sufficiently gradual so as to minimize or reduce to an operationally acceptable level optical power transfer into undesirable modes of the coupled-waveguide system.
The “approach regions” of the joined waveguides (i.e., the regions directly before and after the junction region; may also be referred to as input and output regions) should preferably be adapted to satisfy the adiabatic condition. The waveguides to be joined may typically approach each other at a fairly shallow angle in order to minimize undesirable optical power transfer or optical loss that might result from an abrupt approach. Alternatively, one waveguide may arise from a narrow tip and increase in height and/or width along the length of the other waveguide before reaching its full transverse dimensions. This gradual “appearance” of optical material may be made sufficiently gradual so as to satisfy/maintain the adiabatic condition. Similarly, after the junction region, the waveguides may move apart at a shallow angle, or one waveguide may decrease in transverse dimension(s) until it terminates in a narrow tip. The relative lengths of the approach regions and the junction region will typically depend on the strength of the interaction between the joined waveguides. For strong interaction between the waveguides in the junction region, the junction region might be relatively short, while very gradual approach and separation of the waveguides (and correspondingly longer approach regions) may be required to maintain an adiabatic condition. On the other hand, weaker interaction between the waveguides in the junction region requires a relatively longer junction region to achieve a given level of optical power transfer, but shorter approach regions may be used while nevertheless substantially avoiding undesirable optical power transfer to other optical modes. For a given waveguide type/geometry, it should be possible to achieve a desired level of optical power transfer between the waveguides with undesirable optical coupling maintained below some operationally acceptable level, while minimizing the overall length of the adiabatic optical power transfer device. If a higher level of undesirable optical coupling is tolerable (i.e., operationally acceptable) in a given device, shorter approach regions may be employed in order to reduce overall device size. It should be noted that the approach regions and junction region may not be clearly demarcated, but instead may gradually transition from one to the next. Such gradual transitions are typically necessary in order to satisfy the adiabatic condition.
For achieving substantially complete transfer of optical power between the waveguides, Δβ preferably reaches zero and changes sign at some point within the junction region, after which |Δβ| increases along the junction region. At the end of a sufficiently long junction region (i.e., |Δβ|>>|κ| at z=L; output termination condition), the system mode carrying the optical power has evolved to substantially resemble the isolated-waveguide mode of the second waveguide, and the optical power leaves the coupling region in the second waveguide. The first waveguide may or may not terminate at the end of the junction region or shortly thereafter, provided that such termination satisfies the adiabatic condition. Likewise, the second waveguide may only appear at the beginning of the junction region or shortly before, provided that such appearance satisfies the adiabatic condition.
It is important to note that adiabatic transfer of optical power from the first waveguide to the second waveguide is accomplished without the use of “mode coupling.” In particular, optical power leaves the junction region on the second waveguide carried by the “same” system mode as the system mode that carried the optical signal power entering the junction region on the first waveguide. This occurs since the adiabatic condition dictates that only negligible (or at most operationally acceptable) optical power transfer to other modes has occurred during the transfer of optical power between the waveguides (i.e., the system mode has been preserved by the adiabatic properties of the junction, even though its physical appearance has evolved in transit through the junction region). This behavior is quite distinct from the behavior of mode-interference coupling, which relies upon optical power being carried through the junction region by multiple system modes (usually two) to achieve optical power transfer.
In order to achieve division of optical power leaving the junction region between the two adiabatic-coupled waveguides (having entered the junction region through only one of them), the junction region may be configured so that at z=L the system mode substantially resembles a superposition mode that includes substantial components of both isolated-waveguide modes. Under these conditions optical power in the system mode will be divided into the two isolated-waveguide modes and leave the junction region in both waveguides. For example, a desired fraction of optical power transfer of about 50% (i.e., about 3 dB) may be desirable for implementing an interferometric device. An adiabatic optical power transfer junction may be employed having |Δβ| decreasing to about zero and then remaining near zero over the remaining length L of the junction region. The resulting system modes may have substantially equally weighted components substantially corresponding to each of the isolated-waveguide modes at the end of the junction region, resulting in substantially equal fractions of optical power leaving the junction region in each waveguide. Other fractions of optical power transfer may be implemented by employing adiabatic transverse optical power transfer as required for a specific device.
In contrast to the behavior of mode-interference-coupled waveguides, in which optical power transfer oscillates as a function of the junction length L, the fraction of power transfer for adiabatic optical power transfer is a substantially monotonic function of the distance L, typically closely approaching an asymptotic value after a certain minimum distance (which depends on κ and Δβ) and then remaining substantially unchanged with additional junction region length. This fundamental difference in behavior has a profound influence on the fabrication/assembly/alignment tolerances required for producing transverse-coupled optical components. Briefly, variations in κ and/or Δβ may affect the minimum junction region length required to achieve a desired level of optical power transfer between waveguides, but do not typically affect the asymptotic fraction of optical power transferred. As long as the junction region of an assembled device is longer than the largest such minimum junction region length likely to arise due to fabrication/assembly/alignment variations, then the fraction of optical power transfer in the assembled device will remain substantially unaffected. This is discussed in more detail below, and is an important feature of the present invention.
Frequently the desired objective of an optical junction device is to effect a specific degree of optical power transfer from one optical component to another optical component assembled therewith. Achieving a specifically-desired degree of optical power transfer using mode-interference coupling requires design, fabrication, and assembly of transverse-optical-coupled elements having κ, Δβ, and L kept within tight tolerances (although not as tight as tolerances required for end-coupling, as discussed in reference A8). Variation in relative positioning of the optical elements (affecting κ and possibly also Δβ) causes variation in the “beat length”, and hence the degree of optical power transfer for a given junction region length L (which may typically range between several tens to about 100 μm). For example, a fiber-optic taper segment (diameter 2-3 μm) mode-interference-coupled to a top surface of a dielectric waveguide on a substrate (3-5 μm wide) may require positioning within ±0.5 μm accuracy horizontally and ±20 nm accuracy vertically to keep nominally complete (100%) optical power transfer above the 90% level (0.5 dB level). Such tolerances may be difficult, expensive, and/or time-consuming to achieve, and may reduce device yield, particularly in a mass-production environment. Similarly, the mode-interference-coupled elements must be designed and fabricated sufficiently accurately to yield sufficiently accurate Δβ and κ. Variation in Δβ gives rise to variation in the beat length as well as the maximum degree of optical power transfer that may be achieved. Further discussion of mode-interference coupling, and optical coupling in general, may be found in Fundamentals of Photonics by B. E. A. Saleh and M. C. Teich (Wiley, New York, 1991), hereby incorporated by reference in its entirety as if fully set forth herein. Particular attention is called to Chapter 7 and Chapter 18.
Adiabatic optical power transfer may be exploited to further relax manufacturing tolerances for assembled optical components and devices relative to mode-interference coupling. For example, to achieve substantially complete transfer of optical power from one waveguide to another using adiabatic optical power transfer, the length of the junction region should be made sufficiently long (typically several hundred μm up to perhaps several mm) so that substantially complete optical power transfer occurs for nearly all values of |κ| and |Δβ| likely to arise during fabrication and assembly of an optical device. Manufacturing variations in κ and Δβ would therefore have little or no effect on the substantially complete transfer of optical power between waveguides (in contrast to the situation with mode-interference coupling). For example, in the example given above of a fiber-optic taper segment (diameter 2-3 μm) coupled to a top surface of a dielectric waveguide on a substrate (3-5 μm wide, with a modal index varying over a junction region several hundred μm in length) may only require position accuracy within limits about 3 to 5 times larger than those required for mode-interference coupling.
The techniques and configurations of adiabatic optical power transfer may therefore be exploited for constructing optical devices that include initially separate optical components subsequently assembled together, thereby providing apparatus and methods for transferring optical signal power between optical components that overcome various drawbacks described hereinabove. It is desirable to provide apparatus and methods for transferring optical signal power between waveguides joined by an adiabatic optical junction. It is desirable to implement apparatus and methods for adiabatic optical power transfer wherein fabrication, assembly, and/or alignment tolerances are substantially relaxed relative to end-coupling and mode-interference transverse-coupling. It is desirable to enable passive alignment of the waveguides. It is desirable to provide at least one of the waveguides as an integrated optical component on a substrate. It is desirable to implement substantially adiabatic optical power transverse-transfer adiabatic apparatus and methods that may be compatible with established optical device technologies.
A fundamental problem in the field of fiber-optic telecommunications is efficient transfer of optical signal power between the optical fiber and the optical devices for generating and/or manipulating the optical signal power. Transverse-transfer of optical power may be advantageously employed to transfer optical power between an optical fiber and an optical device through an intermediate external-transfer optical waveguide. It is desirable to provide apparatus and methods for transferring optical signal power between an optical device on a substrate and a transmission waveguide through an external-transfer waveguide optically integrated with the optical device on the substrate, wherein optical signal power is transferred between the external-transfer waveguide and the transmission waveguide by optical power transverse-transfer (adiabatic or otherwise). Optical power may be transferred between the device and the external-transfer waveguide by end-transfer or transverse-transfer (adiabatic or otherwise). An external-transfer waveguide adapted for end-transfer with the optical device may be substantially spatial-mode-matched therewith. The transmission optical waveguide may be the optical fiber (suitably adapted for transverse-transfer) or may be a planar waveguide. Such a planar transmission optical waveguide may more readily enable transfer of optical signal power to/from the optical fiber. It is desirable to implement optical power transfer via external-transfer waveguide apparatus and methods that may be compatible with established optical device technologies. An external-transfer optical waveguide (adapted for optical power transverse-transfer, adiabatic or otherwise, with a transmission waveguide) may be a component optically integrated with an optical device, and may be provided using precision spatially selective fabrication and processing techniques similar to those used to fabricate and process the optical device. Use of such fabrication techniques thereby enables wafer-scale fabrication and precision alignment of many external-transfer waveguide/device pairs in parallel on a single substrate, thereby realizing significant economies of time and cost to manufacture optical devices. It is desirable to enable and/or facilitate substantially simultaneous assembly/alignment of an optical device with two or more transmission waveguides.